1. Field of the Invention
This invention pertains to a microphone/speaker combination which picks up speech vibrations on a unidirectional basis and develops an isolated signal for audio processing which deletes background audio signal. More particularly, the present invention relates to an ear mounted microphone/speaker combination responsive to speech from the user which is substantially isolated from surrounding background noise and which can be used in duplex without occurrence of feedback.
2. Prior Art
The emergence of computer based control systems for virtually every aspect of daily living has re-emphasized the need for an inexpensive system for providing direct voice command capability. At present, most computer systems rely on keyboard input to provide data and processing instructions. This is not only time consuming and limiting in view of physical requirements of being seated with both hands on a key pad, but it limits the efficiency of data input to the typing speed of the user. Indeed, compared with the advanced degree of sophistication of current computer systems, this dominant data entry method of imputing instructions one letter or number at a time seems almost archaic.
This is not to say that the development of voice activated input systems has been ignored. There has clearly been a long recognized need for implementation of a method of voice command which would obviate the keyboard as a computer input device. Numerous problems have remained unsolved, however, and continue to place voice activated systems beyond the financial reach of the general marketplace. For example, the typical level of confidence in voice recognition for commercial software is between 45 to 60%. At this rate, the system is averaging only one out of two words with correct identification. This is hardly adequate for general application. This confidence level is even less when environmental factors are introduced such as background noise, mobility of the user and other unpredictable variables that modify the audio signal being received at the computer.
In typical applications for voice activated systems, one might expect a telephone to ring during dictation, an interruption to occur by another voice having a different voice pattern, the sounding of a horn, clock alarm, or any of thousands of extraneous background noises. Even the shifting of position of the speaking individual can modify the audio signal enough to confuse the computer recognition of sounds. The speaker may simply turn her head to a different direction and create enough variation to thwart voice recognition.
Resolution of these anomalies must be accomplished before voice command systems can be of general use. Current efforts have focused on generation of complex algorithms which sift through the maze of audio signal and attempt to classify and discard background noise. Obviously, it is impossible to completely predict and program every potential superfluous signal that may eventually be detected. Voice variations by the user are even more challenging, because the occurrence of a cold, sinus problems, and variations in voice orientation to the microphone are extremely difficult to classify or quantify. As a consequence, voice command systems are marginal in overall performance and are extremely expensive.
Undoubtedly, part of the problem of voice recognition arises because of the use of conventional microphones as the media for processing voice signals. Voice recognition obviously requires conversion of the audio input to digital or analog signal. Generally, a digital conversion is necessary before comparison with database information is possible. A normal microphone will pick up sounds indiscriminately, thereby adding in background noise which is a primary source of confusion.
Although some discrimination of sound can be accomplished by frequency filters, proximity attenuators, and related devices that group sound characteristics as to common features of frequency or volume, total isolation of desired sounds has not generally been accomplished when the speaker is in a nonisolated environment. Although some applications of unidirectional microphones have provided improved isolation, development of such a totally selective microphone system has long been needed and would provide great advantage in countless applications. This is so with any form of telephone system that is hands free so that the microphone is picking up background signal along with the primary speaker's voice. Visor microphones for cellular car phones are just one of a myriad of examples. TV and stage cameramen, as well as other sound technicians who must communicate quietly while recording or producing video and sound works, require constant communication which functions best in an isolated mode. Switchboard operators, pilots, motorcyclists, workers having on-line communication systems in high-noise environments--all need a communication system which isolates the voice signal from all extraneous sounds.
U.S. Pat. No. 4,588,867 by Konomi addressed some of these problems and disclosed the possible utility of an ear-mounted microphone for discarding air-borne sound waves in favor of bone conduction of the speaker's voice. Aside from technical problems which may have affected the inability of this device to perform adequately, practical considerations also limit the general acceptance of such ear-mounted microphone systems. For example, the Konomi device places a microphone element deep in the ear canal in order for vibrations to be accessed from the bone. This requirement immediately imposes the high cost of having the canal insert portion sized to the unique configuration of the user's ear canal as with a hearing aid. It also adds the discomfort of having an insert in the ear, creating an unnatural and unacceptable sensation which would undoubtedly preclude universal usage of the device.
As a further problem, the Konomi reference may be subject to audio feedback arising from the close proximity of the microphone element and speaker, where a duplex mode is selected. Konomi attempts to reduce such feedback by insertion of sound absorbing barriers between the microphone and speaker portions of the device. Because a total barrier is virtually impossible to develop, there is always some sound which penetrates, generating some feedback signal. Needless to say, the challenge of an acceptable ear-mounted microphone was not resolved with Konomi.
U.S. Pat. No. 4,930,156 by Norris addressed the problem of audio feedback with its disclosure of an ear microphone/speaker combination; however, the microphone was placed outside the ear and was sensitive to ambient noise. The absence of isolation of the speaker voice gave the device only limited acceptance, despite the fact that it offered some improvement over the prior art.
U.S. Pat. No. 4,009,355 by Poradowski, and U.S. Pat. Nos. 3,995,124 and 3,922,488 by Gabr represent a line of directional microphones which are generally referred to as unidirectional systems. These microphones typically provide for opposing openings in opposite ends of the microphone casing. This permits sound waves to enter the casing from different orientations with respect to an enclosed speaker element. If the origin of the sound waves is at a substantial distance from the microphone, the arrival of each sound wave at each of the opposing openings is substantially concurrent. Because this sets up two opposing acoustic waves within the speaker casing, the waves cancel and the sound is basically undetected or unregistered at the speaker element.
This technique has been used to reduce ambient noise and provide some speech isolation for phone operators, camera staff, and other isolation communication systems as designated above. Typically, this unidirectional microphone is positioned within a boom or forward extending support arm that places the microphone immediately in front of the speaker's mouth. It's close proximity to the mouth provides a differential response of sound waves at the opposing openings, based on the time difference in arrival of the wave at each opening. As a consequence, the proximate sound waves do not cancel, giving some selectivity to the near speech, as opposed to sound generated at a sufficient distance that the acoustic waves arrive at the respective openings in the speaker casement at the same time.
Nevertheless, there remains a dissatisfaction with the inconvenience of a boom suspended in front of the speaker's face. It imposes an obstacle that interferes with movement of food or drink toward the user's face, as well as providing a limiting factor to movement which might jostle the microphone from its proper position. Cosmetically, it is obviously unpopular. However, because prior art attempts to place the microphone away from the face and toward the ear give rise to feedback problems with the associated speaker positioned at the user's ear, the boom remains the primary means of isolating speech in a microphone system.
Unidirectional microphone systems have been limited to isolated microphone applications because the combination of a microphone with a speaker element has typically resulted in feedback. Control of feedback in duplex mode for a mic/speaker combination has been the subject of continued research for many years. Although it has been generally recognized that the ear is an ideal support location for a duplex mic/speaker, overcoming technical problems such as feed back and sound isolation has presented a formidable challenge.
What is needed therefore is a microphone that more effectively isolates the voice of the speaker from the environment, yet permits use with a speaker in the same housing without feedback. Such a device should permit use at the ear without insertion of physical structure within the ear canal, with attendant fitting expense and discomfort.